A&O Shearman, one of the world's largest law firms, announced on Wednesday a commitment of $1.1 billion over five years to accelerate its artificial intelligence strategy — with the majority of that investment coming not from technology budgets, but from lawyers' time: every practice group will invest at least 3% of its hours developing specialist AI tools and services for its own area of law.
The announcement, published by the firm and reported by the Financial Times and Reuters, represents one of the largest single commitments to AI yet made by a professional services firm, and it is notable mostly for what the firm is betting on: domain expertise rather than raw model capability. For more context on this story, see our ongoing artificial intelligence updates.
How the $1.1 Billion Commitment Works
Most of the $1.1 billion will be delivered through billable time rather than cash. Each of the firm's practice areas is required to devote at least 3% of its time to building AI tools and services tailored to its field, alongside direct technology spending. The Financial Times reported that partners agreed to the plan this autumn.
Practice heads can decide how to meet the target — assigning dedicated staff full time or spreading the work across partners and associates — and the firm says it will monitor hours across each practice to ensure the commitment is actually delivered. That measurement detail matters: a time-based commitment at a law firm is only as real as the timesheet tracking behind it, and A&O Shearman is explicitly putting the target into the firm's accounting rather than leaving it aspirational.
The strategy also formalizes what had been creeping in informally. This summer, the firm introduced AI usage as a bonus and performance metric for associates, partners and business functions, with weighting that differs between tasks such as individual productivity and the creation of contract precedents and prompts for wider team use, according to the firm's global head of AI, David Wakeling.
Lessons From Four Years of Deployment
A&O Shearman is unusual among large firms in having tested AI deployment at scale for four years. Its conclusion, stated plainly in the announcement: AI delivers the most meaningful impact when teams embed deep legal expertise into systems that can be systemized across a particular area of law — not when a single set of generic tools is applied uniformly across the firm.
The firm says this approach allows premium legal expertise to be delivered more effectively than off-the-shelf tools, and — critically, in its framing — delivers better outcomes for clients. The five-year program is designed to systemize those AI modules across practices rather than chase each new model release.
"This significant investment signals where we intend to take the firm over the next decade," the firm said in its announcement, positioning the plan as a step change in how legal work is designed and delivered and in how the firm trains its lawyers.
ContractMatrix and the Ninefold Surge
The clearest proof point for the strategy is ContractMatrix, a contract drafting and negotiation product the firm developed with Harvey and Microsoft and sells directly to clients, often under novel pricing structures. Each engagement deploys a team of lawyers, data scientists and developers across multiple jurisdictions.
The numbers behind it have moved fast. Client uptake of the ContractMatrix product suite increased ninefold over the last 12 months, according to the firm. Wakeling declined to disclose detailed financials but said revenue is in the tens of millions of US dollars, having risen ninefold over the past year from the previous 12-month period.
The product also recently won a multiyear mandate from a global financial institution in a blind evaluation in which, the firm says, ContractMatrix outperformed the client's own internal data science team, specialist legal AI systems and foundation model providers. For a firm selling AI services against AI-native startups, that evaluation result is the marketing claim it will lead with.
Early-Mover Context and Billing Pressure
A&O Shearman's position in legal AI was earned early. The firm — formed in 2024 through the merger of Allen & Overy and Shearman & Sterling — was Harvey's first customer in 2022, and its 2022 rollout of generative AI was the first by any law firm globally, according to the company.
Large law firms now broadly use Harvey or its European rival Legora, alongside general-purpose tools like ChatGPT and their own internal software, and competition is intensifying over how AI reshapes legal work. Reuters framed the announcement as building on law firms' wider tech investments — with AI incumbency now a differentiator in pitches for sophisticated corporate work.
There is also a pricing dimension. The firm says its fixed-fee income has increased 15% over the past two years, partly attributed to AI, as clients push law firms away from hourly billing toward value-based pricing. Most fees remain tied to the billable hour, but AI-driven efficiency gives firms an incentive to restructure pricing — and gives clients leverage to demand it. A firm that can produce contract reviews in a fraction of the hours cannot indefinitely bill by the hour; committing $1.1 billion to AI is partly a bet that owning the efficiency curve is better than having it forced upon you.
What It Signals for the Legal Market
The commitment sets a benchmark rivals will be asked to match. If a five-year, billion-dollar-plus plan becomes table stakes among global firms, the legal AI market — already one of the fastest-moving verticals for enterprise AI adoption — will consolidate around vendors that can support practice-specific, deeply integrated deployments rather than general-purpose chatbots.
It also sketches what the billable hour may become in an AI-first firm: less a unit of revenue and more a unit of investment, with 3% of every practice's time explicitly reserved for building the systems that will eventually do the work those hours used to cover.
---
Stay Ahead of AIGet the latest AI news, analysis, and breakthroughs — all in one place.
Read more AI news →
